Area-wide river flow modelling for the Thames Estuary 2100 project: Model formulation and assessment

The overall purpose of the Thames Estuary 2100 project is to develop a Flood Risk Management Plan for the Thames Estuary (including London) over the next 100 years, a period long enough that climate change must form an important consideration. Here, the changing fluvial flood risk induced by storm rainfall is the focus of concern. Since the 1960s, daily precipitation in the UK has tended to be more intense in winter and less intense in summer (Osborn and Hulme, 2002). The impact of increases in rainfall on river flows will depend on both the nature of the rainfall and the physical characteristics of the catchment draining to the river. For fast-responding catchments, such as those in impermeable or high relief areas, the characteristics of the specific rainfall event are critical. Such catchments tend not to have the deep soils and permeable geology that lead to the long-term hydrological “memory” of larger lowland catchments. Catchments across the Thames Basin are typical of these lowland catchments and where the longer-term balance between rainfall and evaporation is particularly important. Against this background this report describes and assesses a distributed hydrological model for the Thames Basin that provides estimates of the fluvial flows into the Thames Estuary. The model structure aims to incorporate the spatially-varying effects of soil and terrain on river flows. A single hydrological model, known as the Grid-to-Grid model, is used to estimate natural river flows continuously at a 15 minute resolution on a 1km grid covering the Thames Basin. The rainfall input to the model employs data from a network of raingauges that have been spatially interpolated onto the 1 km model grid. Potential evaporation estimates on a 40km grid across the UK are provided by MORECS. The model is configured spatially using digital datasets representing elevation, slope, soil-type and urban land-cover. Flow directions, required by the routing component of the model, define the drainage direction of water from one model grid-cell to the next. These have been derived from hydrologically-corrected higher resolution (50m) digital terrain data. Output from the Grid-to-Grid model consists of a time-series of grids of estimated river flows across the Thames Basin at 1km resolution and for a 15 minute time-step. The model is also able to provide 15 minute gridded estimates of model internal variables such as soil-moisture and surface- and sub-surface runoff. The accuracy of the Grid-to-Grid model river flow estimates is assessed at 34 gauging-station locations across the Thames Basin during a 5- year period from 1997 to 2001. The Grid-to-Grid model performs well for a wide range of catchments across the modelled domain, particularly where the catchment response to rainfall is considered natural. It is important to note that the model is unable to simulate artificial influences on river flows such as effluent returns and groundwater abstractions. Not all the gauging stations provide accurate flow estimates: the model assessment has taken this into account, lending greater weight to model performance at reliable gauging stations. Inaccurate estimation of spatial rainfall over the Thames Basin is another potential source of error: quality-control of both the rainfall and flow data has attempted to minimise both sources of error. The spatial consistency provided by a single model with a single set of parameters allows a water-balance to be maintained for the Thames Basin, with the result that, aside from evaporation, all rainwater falling in the catchment will eventually reach the estuary. Ongoing and future work will use the Grid-to-Grid model to assess future changes in river flows in the Thames Basin. The model will employ as input ensembles of rainfall and potential evaporation obtained from the Hadley Centre’s 25km Regional Climate Model (RCM) for different climate change conditions and futures. Flow scenarios will be developed for flow in the main Thames and twelve major tidal tributaries using the RCM to provide dynamic downscaling from the Global Climate Model. The study will produce estimates of spatial changes in flood peaks of given return periods across the Thames Basin including tributary inflow points to the tidal Thames.